A CONTEXT AWARE BASED PRE-HANDOFF SUPPORT APPROACH …

Journal of Engineering Science and Technology Vol. 10, No. 6 (2015) 743 - 764 © School of Engineering, Taylor’s University

743

A CONTEXT AWARE BASED PRE-HANDOFF SUPPORT APPROACH TO PROVIDE OPTIMAL QOS FOR STREAMING

APPLICATIONS OVER VEHICULAR AD HOC NETWORKS – HOSA

K. RAMESH BABU*, A. THANGAVELU

School of Computing Science and Engineering, VIT University, Vellore, India

*Corresponding Author: [email protected]

Abstract

Large variations in network Quality of Service (QoS) such as bandwidth,

latency, jitter, and reliability may occur during media transfer over vehicular

ad hoc networks (VANET). Usage of VANET over mobile and wireless

computing applications experience “bursty” QoS behavior during the execution over distributed network scenarios. Applications such as streaming

media services need to adapt their functionalities to any change in network

status. Moreover, an enhanced software platform is necessary to provide

adaptive network management services to upper software components.

HOSA, a handoff service broker based architecture for QoS adaptation over

VANET supports in providing awareness. HOSA is structured as a middleware platform both to provide QoS awareness to streaming

applications as well to manage dynamic ad hoc network resources with

support over handoff in an adaptive fashion. HOSA is well analysed over

routing schemes such as TIBSCRPH, SIP and ABSRP where performance of

HOSA was measured using throughput, traffic intensity and end to end delay. HOSA has been analysed using JXTA development toolkit over C++

implemented classes to demonstrate its performance over varying node

mobility established using vehicular mobility based conference application.

Keywords: Quality of service, Streaming media, Handoff, VANET, Middleware

policy management, Mobility, Context management.

1. Introduction

“Bandwidth on demand” streaming applications depend on underlying

communication network infrastructure to provide access to user intensive services

and utilization of resources. Ideally these applications do not concern about the

networks used but focus only on the service functionalities being provided, which

744 K. R. Babu and A. Thangavelu

Journal of Engineering Science and Technology June 2015, Vol. 10(6)

Nomenclatures

Ci Contour

D Distance between Thmin and Thmax

ST Signal strength

Thmax Maximum handoff time

Thmin Minimum handoff time

Tmax Maximum time to expire

Tmin Minimum time to expire

Abbreviations

AODV Ad hoc On Demand Distance Vector Routing

API Application Programming Interface

GPRS General Packet Radio Service

GPS Global Positioning System

HAND Handoff Module

HOSA Optimal QoS Middleware for Streaming Media over VANET

IETF Internet Engineering Task Force

IrDA Infrared Data Association

JXTA Juxtapose

QoS Quality of Service

TC Traffic Class

TOS Type of Service

VANET Vehicular Ad-hoc Network

WLAN Wireless Local Area Network

is complex to support in practice. Large variations in network quality of service

(QoS) (e.g. bandwidth, latency, jitter, reliability) may occur during media transfer

over dynamic, highly volatile ad-hoc networks which lead to service degradation

and increase the faulty nature of system.

Communication over mobile networks is established based on device

portability [1] and wireless network connectivity [2]. The end user’s terminal

highly vary in terms of processing capabilities, input and output capacities, energy

consumption, and networking technologies including error rate, signal fading and

interference, etc. User mobility on a mobile or wireless device leads to consistent

change of location, environment, network operator or service provider, and access

networks. Issues such as handoff during a complete session add to the source of

network variations in terms of session transfer, end to end delay, packet loss, jitter

which contribute to latency factors of network.

In this paper, an optimal QoS scheme with resource reservation policy approach

for VANET is discussed, with support over handoff reservation over a call session

based on optimal service quality metrics, as well as consistent monitoring and

controlling with binded quality during dynamic runtime environment. The objective

of the model is to explicitly deal with the coexistence of ad-hoc networks built by

multiple devices supporting distinct technologies and aiming to exchange media

streaming messages. Based on this model, and assuming that in general, nodes are

not able to directly communicate one with each another, a set of resource awareness

A Context Aware Based Pre-Handoff Support Approach to Provide Optimal . . . . 745


services were devised to establish communication path among nodes through

intermittent available nodes. The proposed set of services [3, 4] consider the

policies for sharing network resources such as bandwidth, route capacity

management along with service profile preferences, thus allowing to tailor the

network services. Resource awareness services are layered under a set of

communication services in a middleware architecture enabling communications

among nodes belonging to distinct ad-hoc networks.

From the focus of research the following issues need to be addressed

• To analyse the aspects of providing QoS over VANET with support for pre handoff.

• To design a middleware which needs to provide support over QoS, resource

management and route control during a call session.

• Suggest mechanisms to handle handoff with optimal utilization of peer node

resources and satisfy end-to-end QoS with minimal response time delay over

media streaming applications.

To provide quality of service (QoS) with adaptable end-to-end delay, a

middleware approach is proposed with support for handoff mechanisms. On

analysis it has been identified that the current generation of commercial-off-the-

shelf routing protocols lack in providing adequate QoS support and handoff

brokerage in any changing, dynamic environments.

This paper contributes to the study of adaptive middleware by first suggesting

how priority and resource reservation with network QoS management

mechanisms can be coupled with the existing standard off-the-shelf distributed

network, OS using distributed object computing framework to support dynamic

run time applications with stringent end-to-end real-time requirements. The paper

also discusses on the results of experimentation and validation activities being

conducted to evaluate the objective of QoS and handoff mechanism over

heterogeneous OS, network, and middleware capabilities.

Current QoS models with handoff support functionality such as TIBCRPH [5],

ABSRP [4], and SIP [6, 7] are primarily oriented towards supporting QoS aspects

in network layer and transport layer. But providing an end-to-end control from the

low-level network parameters such as bandwidth, packet delivery ratio and

transmission rate, up to application layer requires a transparent setup to support in

QoS, termed as HOSA as shown in Fig. 1.

HAND is a handoff module which works as a middleware in

communication with multiple VANET nodes on mobility. HAND provides

support to optimal QoS with effective resource reservation and utilization

among nodes on utilization. HAND defines network components as objects

and provides a distributed binding among objects since there is an increasing

need for network usage aware applications with focus on network

performance and quality. The service and components should be more

intelligent of network in order to adapt to various environments. On the other

hand despite of the differences in the functionality of the applications, the

main networking services can still be abstracted to common and unified

interfaces to be used by diverse applications. Adaptation functions need to be

rationally distributed into both specific applications and a general system

platform, and then implemented with distinct mechanisms.



Fig. 1. HOSA End-to-End QoS Architecture.

HOSA (an Optimal QoS Middleware for Streaming Media over VANET) is

middleware architecture for adaptive applications based on network resources and

service awareness. The architecture works on physical, data-link and network

layers of mobile devices. HOSA provides flexible network support to adaptive

network applications with a set of functions and interfaces. The overall

architecture of HOSA in this paper lays emphasis on the methodology of

delegating the application and network components to form context aware and

support in providing adaptive networking services to network-aware applications.

The rest of the paper is organized as follows: Section II presents the literature

review on related handoff QoS architectures. Section III discusses on the HOSA

architecture and the context issues adopted in HOSA as well the methods to

realize context awareness. Section IV discusses the realization of adaptive

network supports and the utilization of them in the adaptation of application.

Section V concludes the paper.

2. Review of Literature

Singh et al. [8] had implemented GPS assisted low latency handoff scheme for

802.11 based wireless networks which adopts the longitude and latitude

coordinates of its neighbouring mobile nodes during handoff. Even though this

approach may support handoff with higher throughput this approach consumes

high energy consumption since the mobile nodes are equipped with GPS and

follows ADOV protocols which uses multiple request (REQ) and reply (REP)

procedures to establish handoff.

Subramaniam et al. [9] modelled QARS as a set of high speed vehicles on

straight highway in which any vehicle can establish connectivity with other

vehicles which are travelling in variable directions of its motion. Vehicles which

are within the range can communicate or help in forwarding the data to be

transmitted. QARS uses the forwarding optimization which works in high

contention scenario.

A service discovery approach for vehicular Ad-hoc networks - ABSRP [4]

model adopts vehicle to vehicle communication without using RSUs as



forwarding nodes. In this approach an intermediate node itself acts as routers to

support in vertical handoff and session establishment. ABSRP adopts mobile

nodes as vehicles in a city environment and adopts case studies on vehicular

safety environment.

In [10], the authors had adopted neighbour graph procedure which

dynamically captures pre-positioning of mobile nodes with support on service

context which ensures that mobile node supports multiple services under low

mobility conditions. An accreditable work on scheduler management [11]

proposed adopts packet scheduling and the resource mapping algorithm, which is

based on a cross-layer design. This algorithm suggests that the scheduler is aware

of both the channel at the physical layer as well the queue state at the data link

layer information of achieve proportional fairness while maximizing each user’s

packet level QoS performance. This algorithm strategies packet flow but does not

effectively support in QoS update.

Mishra et al. [12] had carried out an empirical analysis of 802.11 with support

for MCA layer during handoff process which observes that the handoff latency

enthrals with significant impact on degrading the QoS performance for real time

streaming and conference applications. Hosseini et al. [13] work focus on

controlling and avoiding packet flooding in MANET, which uses Cluster Based

Routing Protocol (CBRP). This protocol reduces the load of traffic intensity on

network by minimizing the communication messages as advertisement from

application layer to routing layer. CBRP focuses on only distribute the active

services in the network between clustered nodes. The experimental result show

cases its behaviour with minimal delay such that this method does not add any

extra overhead to the network.

Alexandros et al. [14] proposed handoff mechanism with seamless service

continuity which provides handoff in heterogeneous environment. This approach

enables consistent QoS support in an integrated system with multiple access

technologies. The work also focuses on network context information which needs

to be considered during transition between distributed heterogeneous systems

with fast handoff for the support of seamless service continuity. Although the

mechanism lags in energy consumption, but basically it lags in middleware

support which is considered as major work of HOSA.

This work also considers routing schemes such as Mobile IPV6 [15], SCTP

[16] and UMTS [17] mobility protocols where the applications does not consider

the mobility status of a node. The authors proposed a middleware solution, MUM

[18] which supports the phenomenon of context-aware handoff management to

avoid service interruptions during both horizontal and vertical handoffs. This

approach exploits the visibility of wireless sessions available along with handoff

implementations (handoff awareness), service quality requirements and handoff-

related quality degradations (QoS awareness). MUM [15] approach provides

solutions for handoff prediction, consistent multimedia transfer but with higher

degradation quality of media streaming issues due to improper data buffering.

Wang et al. [5] had proposed TIBCRPH which adopts the network traffic

infrastructure and clusters the node in network with support over routing with

Handoff capabilities. This approach adopts the vehicular density with its speed

(mobility), distance between vehicles and time taken to cross between vehicles.

Even though cluster analysis provides support in QoS but pro-active support to



handoff does not exist as handoff metric analysis is not effective in

implementation. SIP [6] is a session management protocol supported as a standard

by IETF for invoking multiple sessions over a communication scenario. SIP had

been implemented over VANET [19] for maintaining a robust handoff over

multiple intermediate nodes but this protocol does not claim to provide an

efficient QoS mechanism. Musolesi and Mascolo [20] formulated on Context

aware Adaptive Routing (CAR), which is a prediction based routing protocol with

support over delay tolerant ad-hoc networks. A source node willing to send a

message to a destination adopts Kalman Filter prediction with multi-criteria

decision making theory to select next hop for message forwarding. Any node with

high mobility is considered a good carrier of data since it follows multiple nodes

during its transmission. Similarly the existing colocation pattern indicates that the

node will meet the recipient again during next transmit.

Wang and Qian [21] proposed a mobility handover scheme called as MHVA

for IPv6-based vehicular ad hoc networks. This scheme adopts mobility handover

mechanism which demands completion of mobility handover operation in

network layer before similar operation is carried out in link layer is performed.

Yu-Doo Kim at al. [22] work on comparison of routing protocols analyses the

performance of IEEE 802.11 and IEEE 802.15.4 standards for sensor and ad hoc

networks. Though this work does not carry much importance on handoff issues,

this work suggests various possibilities of working on identifying QoS.

Johann et al. [17] has discussed the development trend of VANET in the

future, which addresses the key issues of mobility management in VANET, as

well analysed the issues behind the existing technology which also focuses on

Handoff issues in heterogeneous environment. Based on analysis and survey of

literature work being carried out, HOSA is proposed.

3. HOSA

The need for middleware architecture with support to QoS with variable

services and network resource utilization over VANET is a highly demandable

research challenge among the erstwhile research issues. HOSA is modeled

primarily as a middleware with an aim to provide optimal QoS over VANET

and other network architectures.

3.1. Middleware architecture and functionality

HOSA Middleware Architecture together with Policy Controller, Service

Manager and other components in the execution environment is shown in Fig. 2.

The execution environment includes the adaptive application as the consumer of

the HOSA’s networking services, and underlying infrastructure that serves as the

necessary supporting modules.

HOSA considers all nodes as VANET mobile / wireless nodes which

communicate with each other as heterogeneous nodes or clients. The

phenomenon of VANET network device monitoring includes core ad-hoc

components such as adapter type, road side units in use, gateway

interconnectivity and beacon signaling rate. To control the node on mobility,



the basic protocol entities such as TCP, UDP and IP uses interfaces for

management of multi-deterministic device drivers, protocol stack arrangement

and routing information update. Nodes communicate with each other using

Hello protocol [23, 24] and establish routing path such that a session is

maintained and consistently updated over a period of time using query of

context information. The node and service contextual information is being

locally positioned in an end host and globally accessible as a central server.

Section III discusses context and context awareness in more detail.

HOSA middleware is the software platform which is defined above operating

system and other network resource infrastructure, providing adaptive network

connectivity management to upper system modules and network applications.

The infrastructure shown in Fig. 2 comprises of components under the

monitoring, control, and management of resources with extended services towards

handoff brokering and providing QoS functionality for any service in use.

HOSA’s distributed node caller uses QoS policy manager to express adaptation

rules, which enables an easier way to handle utilization of adaptive services

functionality provided by HOSA, in which the application needs only to represent

the requirements by policies. The detailed processing of the adaptation demands

is managed and controlled by Connection Controller without the concrete

concerns of service in use. Connection controller forms the management core of

HOSA middleware. The service in use is realized by the creation and maintenance

of Service Event Handler.

Fig. 2. HOSA Middleware Architecture Space.



The service based adaptation function is handled by Adaptation Selector and

Adaptation Executor modules, which selects the service and used connection

controller to adapt to network. The adaptation of a provided service can be

realized through multiple serial stages, including adaptation triggering, approach

selection and adaptation execution. Adaptation mechanisms are first triggered by

some specific context according to the predefined matching criteria. Then a

decision should be made on which adaptation approach will be used. Finally,

service adaptation can be achieved by automatically or manually executing a

command and/or changing the external behaviors (and possibly internal states) of

an entity that provide the service. HOSA provides the adaptive network

management services with a set of APIs to the upper network-aware adaptive

applications. From HOSA’s point of view and as the caller of the HOSA’s

functions, such applications are mainly end service applications but can also be

other system level platforms at middleware layer e.g. file access system.

The main feature of the HOSA architecture is the clear partitioning of the

whole adaptive functionality into different levels, in which each level only takes

care of the functions that are most suitable to be concerned by it. The adaptation

of the end application is separated into the application layer and the middleware

layer. All the network adaptation mechanisms are abstracted and placed onto the

HOSA middleware level, since the monitoring and control of network resources are

mostly convenient to be implemented at this level. Semantic oriented adaptation

mechanisms are then left to application level. This is due to the fact that application

knows the content and the media that it consumes and processes the best.

3.2. Mode of operation

The VANET terminal node mobile host can be equipped with several network

interfaces such as Bluetooth, IrDA, WLAN, and GPRS and the availability for

each access technology dynamically changes due to mobility. Network

performance may also be greatly fluctuant in terms of packet loss, latency, and

bandwidth as the result of the wireless connection and handoff. Hence HOSA also

updates context information such as energy consumption, user preference, and

cost incurred for providing a service consistently. As a result, functions in

terminal HOSA focus more on the adaptive management of networking resources

inclusion of application level adaptation. In particular, the adaptive management

of multiple communication channel or connection is the main task of HOSA

terminal node. To manage connection channels is core functions of HOSA, which

is realized by connection controller. HAND Controller functionality is defined

which establishes hand off procedures between set of communicative peers

engaged in session over a service.

3.3. HAND – Handoff Operation

HAND defines a logical connection between set of communicative peer entities,

where a source VANET node initiates a Route Request and the destination node

replies or the request. HAND component of HOSA supports both standalone

operation mode and collaborative operation mode, over the set of VANET nodes

engaged in communication.



HAND defines a wireless / mobile system with identical concentric circles of

zones or contours as shown in Fig. 3. When a node moves from one contour

division to another contour with a speed V at an equidistant D, then mobility

range of nodes is updated at intervals of 50 ms consistently.

The signal strength (ST) of node helps to identify its “nearness” property and

its variable distance provide the speed of node on “mobility”. Any nodes can

engage in handoff based on its equidistant property and the defined threshold to

complete the handoff process. When a VANET node in contour region A initiates

a handoff request whose signal strength is good, then another node in contour

region-2, with an average signal strength can reserve for handoff process to

forward. When node leaves C2 zone, its initiates the handoff since node is in Thmin

zone while when node moves away from C2 to C3 zone then handoff initiation

starts such that it can complete the process within Thmax. The distance between

Thmin and Thmax is D. Selection of Thmin and Thmax is based on the relative signal

strength between set of nodes as well over the range of distance associated with

intermediate nodes.

Fig. 3. Execution of HAND Module.

Handoff initiation algorithm given below in Algorithm -1 creates and controls

multiple handoffs for vehicular node over IEEE 802.11p networks. The basic

principle of this algorithm defines and updates the status of the neighboring

VANET node at every 50 ms through the set of switching or forwarding nodes.

Selection of the forwarding node is discussed in Algorithm-2 where the switching

center maintains the current location and status of VANET node as well as other

nodes related to source node. For each VANET node the corresponding Thmin and

Thmax values are gathered and updated, where Thmin defines the minimal time limit

to initiate a handoff and Thmax defines the maximum time allowed to complete the

handoff. Each VANET node engaged in communication is updated on the

information pertaining to next nearest VANET node with help of source node

which acts as the switching center. The set of sessions and their frequencies are

defined as Ci where i can vary from 1 to n.



Algorithm-1 : HandOff_Initiation ( )

Begin {

Pkts_transmit = 0 // has no packet to transmit

set Session Ci=1 // determine next contour to be selected

if (Pkts_Transmit_Flag=1) and Conn_Quality <=Thmin or > Thmax

{

Set Sleep_Period= Thmax- Thmin // acceptable pause period

call Send_req(Node[i])

call buffer(pkts) // reserve buffer in node ‘i’

Pkts_transmit = 1 // has no packet to transmit

while (!Pkts_transmit)

{

Transfer_Data(Node[i],Node[j])

}

clear buffer

}

select Alternate_Opt_Path // select another alternate path for handoff

call Send_Req(Node[i-1]) // request to another node

NextSession Ci=i+1 // use another session

}

The basic phenomenon of handoff initiation algorithm carries out scanning

and discovery of every neighboring nodes to determine the optimal node engaged

using parameters such as Bandwidth in use, Signal Strength, Bandwidth required,

and Service priority as shown in Fig. 4. Identification of an optimal VANET node

to engage in handoff is such that the selected node is being reserved for the whole

session until the service is completed. This approach eliminates unnecessary

scanning of other VANET nodes to be selected which are far away from the

destined source or destination node. When a node does not receive any data or

packets from its neighboring node then the status is updated by sending a probe

request signal or Hello protocol. The operation of HAND handover initiation

mechanism enables the node to enable multiple decisions making Instead of blind

handoff between the nodes.

Fig. 4. Handoff Request and Handoff Negotiate.



Algorithm-2: HandOff Negotiation ( )

{

while (Status.node[i] = TRUE )

{

A: If (node[i].bStr >Thmin ) // call handoff process

{

Node[i].bndWidth = bndWidth_Update ( ) // bandwidth in use

B: If (node[i].bndReq > = bndReq ( )) // bandwidth required

{

Sid = AssignNode (node[i]) // Session Identifier

}

else

goto B

goto A

}

The handoff process is initiated and carried out over a Negotiate frame

shown in Fig. 4 which communicates with each node neighboring and

designates a node as optimal node for carrying out handoff process. On

negotiation (Algorithm-2), the node whose status is active enough to carry out

the handoff within the TimeToExpire limit then handoff process can be effective

to contribute to the effective QoS. The TimeToExpire field can be defined on

metrics such as priority of Service in use, distance ‘D’ and ‘Tmin’ or ‘Tmax’

metric of a node in a zone to establish handoff and available bandwidth to

complete the process.

4. Context Management in HOSA

HOSA is middleware architecture for supporting optimal handoff transfer over

context-aware network setup which is dependable on network, user and service

applications in use. Two components are related to context awareness, i.e.,

connection monitor, policy manager which supports service based on policy as

per user profile and service adaptation interface, hence binds the node to user

requirement and service in use.

4.1. Service context aware

HOSA provides adaptive applications with the capacity of network awareness

through the information provided by the Connection Monitor module, by

considering the service aware rich context information while on execution.

Additional context information of service can be specified based on time and

event activity. Such information can be accessed through interfaces by connection

controller and applications. HOSA also contributes to the service content in use

by registering the current network information.

The service manager frame as shown in Fig. 5 organizes the service the

required bandwidth to support based on their type, and QoS specifications binded

to their policy list.



Fig. 5. Service Manager Frame Adopted in HOSA.

4.2. Contextual information

Context can be a well-recognized field or object whose tag or context data is

manipulated, based on an activity or event instantaneous. Multiple context based

information is being maintained by system related to a connection management,

which includes:

[i] Static contextual network information: Theoretical network capacities of

network interface (type, typical bandwidth, cell size, handoff latency, power

consumption, user speed, simultaneous user number of the access point),

operator information. The static features allow a raw comparison between

interfaces obtained as measurement during runtime performance through

each interface.

[ii] User profile contextual information: contextual information related to

network management, based on user preference which includes priority of

user interface, interface selection policies. User policies help to balance all

the factors for the interface selection by user. Application policies are also

part of user policy specified by each application which opens a channel for

its traffic flow.

[iii] Network Context information: Contextual information of network such as

energy status, node speed, pause time within a location, distance between

nodes and infrastructure access point are updated and maintained

consistently. This information is highly critical for the adaptive management

of network resources.

For example when the velocity of an end host is too high, the channel should

select and maintain the best connection among the optimal connections available,

and then reduce the quick switch between the two accesses.

4.3. Service adaptation

The adaptation mechanisms for the network management in HOSA are mainly

realized by policy manager and connection controller. In particular as the object

model shown in Fig. 6, the kernel of HOSA is the connection controller acting as

a coordinator and executor.



Fig. 6. HOSA Connection Controller and Policy Manager.

5. Policy Manager and Connection Controller

5.1. Policy manager

HOSA employs a policy mechanism to ease the adaptive management of network

resources. Applications can just express their adaptation requirements with

policies when they open new channels, and so totally disregard the detailed

execution of adaptive mechanisms at the HOSA. A policy denotes the criteria for

the selection of the best current network interface. Then the connection controller

maintains each channel according to the corresponding policy. An application can

also explicitly control the channel with exposed channel control interfaces.

A policy can be either static or adaptive. A static policy explicitly declares the

network interface to be used. An adaptive policy is used to describe the access

selection rule for one particular type of traffic flow. An adaptive policy can be

represented by (traffic class, logic conditions, and weighted factors). Traffic class

(TC) can take any value defines in Type of Service (TOS). Logic conditions are a

series of comparison expressions connected by logic operators. Weighted factors

are a set of 2-tuple (factor, weight).

Sub-policy for achieving optimal QoS over throughput can be defined as:

(THROUGHPUT, (bandwidth>100kbps) and (delay<5s), (cost, 0.2), (trff_load

<0.6), (pktdrop_rate< 0.6),( Service_Use = Audio))

Here the expected bandwidth in use should be more than 100 kpbs with an

acceptable delay of less than 5 seconds or 500 milliseconds and an acceptable

packet load percentile value of less than 6 and variable packet loss between 0.3

kbps and 0.6 kbps.

To define a content delivery over VANET, end to end delay plays a vital role

towards achieving an optimal QoS. Policy can be defined for END TO END

DELAY as a parameter for achieving optimal QoS for variable services such as



(OPTIMAL_QOS, (bandwidth_in_use < 100kpbs), (delay < 10ms),

(traff_load>05), (pktdrop < 0.6), (Service_use=Content))

Similarly to achieve an optimal QoS over highly variable node density, the

following sub policy can be applied

(OPTIMAL_QOS, (bandwidth_in_use < 135kpbs), (delay < 10ms), (node_speed

< 5mts), (traff_load>10), (pktdrop < 0.4), (Service_use=Content))

Policy manager is used by applications to supervise policies including policy

creation and elimination. Policies are then accessed by connection controller during

the channel operations. The evaluation determines which interface is currently the

best to a policy can be done both periodically and immediately when special events

happen. Some of the application policies may conflict with user preference policies

stored previously in the local context server. It has been assumed that user

preferences always have the highest priority, which is the default case.

5.2. Connection controller

Connection controller is the core component of the HOSA for the final realization

of the network management adaptation. By acting as a coordinator and executor,

it controls the activities of local HOSA components such as policy manager and

connection monitor.

The functions of the connection controller are to maintain the information and

manage the channel. Connection information maintained by connection controller

includes both local information and global information. For local information,

connection controller maintains the lists of the references of all the interfaces,

channels and policies, along with the mappings. The lists and mappings are

continuously updated in case of any specific event (e.g. a channel has switched

the connection under using or a new channel is opened with a new policy).

Connection controller maintains global information of the end host the

controller updates the global network contextual data. This update is performed

periodically or when any related event (i.e., changes of numbers and addresses of

interfaces) happens or when a specific event is identified.

5.3. Channel maintenance

Channel is defined as the logical link between physical application components

which are located in different network devices such as terminal or network

connected device being either uni-directional or bi-directional, over a specific type

of connection to transfer data between nodes. The connection established for each

channel can be dynamically updated, while its context values remains unchanged

hence the service continues without any modification. The service may explicitly

control each channel if necessary, through the control of policy manager.

Connection controller periodically re-evaluates the mappings between an

interface and each channel according to the policy used for each channel.

Moreover, the re-evaluation is also immediately done when special events (e.g.

interface up or down, channel opened or closed) occur. If, according to the policy

a better interface is found, then the connection controller initializes a channel



switching session. The session needs the cooperation between controller peers

through the signaling channel, as shown in Fig. 5.

Essential parameters that should be experimented include:

1) Network size measured in the number of nodes

2) Network connectivity the average degree of a node (i.e., the average

number of neighbors of a node)

3) Topological rate of change the speed with which a network's topology is

changing

4) Link capacity effective link speed measured in bits/second,

after accounting for losses due to multiple

access, coding, framing, etc.

5) Traffic patterns protocol effectiveness in adapting to non-

uniform or bursty traffic patterns

6. Experimental Approach

The outdoor routing experiment took place on a rectangular athletic field

(measuring approximately 200 (north-south) by 350 (east-west) meters. Since the

athletic field was distant from the campus wireless network, this can as well help

to reduce the potential interference. The traffic generator on each mobile node

generated packet streams with a mean packet size of 1200 bytes (including UDP,

IP and RTP headers), a mean of approximately 5.5 packets per stream, and a

mean delay between streams of 15 seconds. These parameters generated an

approximation of 423 bytes of data traffic (including UDP, IP and RTP headers)

per node per second, with a modest traffic volume, but corresponding to the

traffic volume observed during trial runs as one of a prototype media streaming

applications [3]. The algorithms are implemented using C++ class based event

controllers which share a core set of API. These classes include the event loop, as

well as unicast and multicast, routing, and logging support.

6.1. Hardware platform

Experiments were conducted using differing IEEE 802.11 supported initially on

set of 25 WiFi nodes as shown in Fig. 7. The experiment was re-defined and re-

executed with 30, 35, 40, 45 and 50 nodes. The nodes were configured to control

and work on ad hoc routing algorithms discussed in Section-2. WiFi cards can

transmit at variable bit rates, the nodes should also auto-adjust the bit rate

depending on the observed signal-to-noise ratio, and arrive at a consistent rate for

all the nodes in network. The nodes are implemented in "ad hoc mode" setup in

which the transmission rate was fixed at variable rate of 2 Mb/s to 200Mbps such

that the channel can select the setup automatically. Specifically, the setup adopts

Lucent (Orinoco) firmware version 4.32 and its proprietary ad hoc "demo" mode

suggested by Lucent.

To ensure consistency with multiple series of ad hoc routing experiments "demo

mode" is adopted. Fixed rate of data transmission is adopted to analyse the routing

results. The multi-rate capabilities over the demo mode propose to use variable bit

rate traffic. Each node enables the wireless communication channel through the

serial port to support accurate distance between nodes throughout the experiment.



Fig. 7. Experimental Test-bed.

The algorithms are implemented in JXTA [25] which share a core set of

classes to support event loop, unicast / multicast routing, scheduling and logging

support. With these four key features, algorithm-specific code is confined to the

packet handler classes that process incoming control and data packets, the timer

handler classes that process timed actions (such as route expiration), the logging

classes that log algorithm events, and utility classes that serialize and un-serialize

the control packets.

The routing algorithms work in synchronization with a traffic generator which

runs on each node, and sends a sequence of packet streams to randomly selected

destination nodes, as part of experiment. Each stream contains a random number

of packets of a random size. Gaussian distributions determine the packet numbers

and sizes, along with the delay between streams and packets, while packets are

uniformly distributed between the source and destination nodes with intermediate

handoff nodes. The GPRS which runs on each node, reads and records the current

node position from the other attached GPRS unit, as well broadcasts the beacons

which contain the source node's position (as well as sequence-numbered positions

that it has received from other nodes).

6.2. Performance analysis

The performance of HAND module was observed along with QoS analysis of

streaming service. In Figs. 7 and 8 the handoff process between four nodes under

communication is explained, where nodes S and R send data and receive data

respectively, while HF1, HF2 and HF3 are three nodes, which are consistently

“on mobile” and act as forwarding nodes to transfer data between the S and D but

variable distances. The data transfer rate between the sender and the receiver with

the help of forwarding nodes show gradual decrease in traffic intensity and packet

transfer rate as shown in Fig. 8.



From the observed phenomena, it was identified that an average of 12% to

18% of packet drop was noticed between the S nodes to HF1 node. Similarly an

average of 15% to 19% of packet drop was noticed between the receding HF1 to

HF2 node, while an average of 18% to 22% of packet drop was noticed between

HF2 to HF3 or D node.

Fig. 8. Execution of HAND Module over 4 Node Communication.

Streaming services which uses log and trace files to account for the effect of

traffic sent and received based on available bandwidth is consistently estimated.

Figure 8 explains the traffic intensity generated over a streaming service for

varying number of mobile nodes where HOSA is being supported with HAND.

Any increase in number of VANET nodes increases the traffic intensity hence

the bandwidth being used and network load. Hence streaming services could

either account for their own traffic explicitly, or operating systems should

support in scheduling methods to send packets at precise intervals without

supporting the process.

Figure 9 supports in understanding the intensity of traffic generated for a

specific service in use. It could be observed that when number of nodes are

between 20 to 25 the traffic intensity generated varies between 80 to 200 bytes for

HOSA approach, where as it shows an abnormal increase of 560 bytes to 700

bytes for TIBCRPH approach

Fig. 9. Traffic Intensity over Increase in Number of Nodes.



Figure 10 explains the throughput of HOSA compared with TIBCRPH,

ABSRP and SIP schemes. The throughput of HOSA with supported pre-handoff

mechanism shows an increase in packet delivery for a fixed set of 25 nodes under

mobility. The graph execution time had been gathered from 100 ms to 600 ms,

considered as a long duration of execution window normally for any experiment.

HOSA converges to optimal local minima due to packet transfer at each time

interval. The experiment is also carried out over varying node density as 25, 50,

75 and 100 nodes. At each experimental run it was observed that HOSA

performed better in throughout compared to TIBCRPH, ABSRP and SIP as

shown in Fig. 11.

Fig. 10. Throughput Observed.

Fig. 12 shows the packet delivery ratio (PDR) at varying intervals of time for

increase in node capacity. HOSA supports higher PDR compared to SIP or

TIBCRPH schemes. The PDR varies from an average of 0.7 percentile whereas

both SIP and TIBCRPH shows abnormal variance in PDR and low delivery rate

with supported handoff components.

The throughput analysis of HOSA is extended over varying node density to

understand behavior based on traffic intensity or load. The experimental run is

carried out over 5, 25, 50, 75 and 100 nodes using ns2 [26] and VanetMobiSim

[27]. Simulation run is adopted since implementing 100 VANET nodes on a real

time setup is highly complex as well demonstrating QoS behavior of HOSA in a

real time setup and simulation setup is highly demanding.

Figure 11 shows the throughput analysis for streaming content executed on

varying node density for HOSA, ABSRP and SIP. Performance of HOSA was

comparatively better than ABSRP and SIP, since SIP consumes more bandwidth

for session establishment and forming the route. The phenomenon of handoff is

improved in HOSA than compared to ABSRP, hence HOSA outperforms.



Fig. 11. Throughput Analysis on Varying Nodes.

The packet drop ratio as shown in Fig. 13(a) is low in case of HOSA due to fast

pre-handoff support module which determines the intermediate node to handoff and

service on demand. The routing schemes converge to higher packet drop when the

nodes varies between 200 to 250 primarily due to an upsurge in traffic intensity,

while SIP show an increase in packet loss compared to other schemes. Fig. 13(b)

shows SIP to have minimal end to end delay compared to HOSA. Performance of

HOSA for streaming services with multiple sessions maintained between variable

users is better compared to TIBCRPH, but SIP invokes buffer at all intermediate

nodes to perform an end to end delay achievable over multiple sessions.

Fig. 12. Observed Packet Delivery Ratio.



Fig. 13. HOSA

(a) Packet Drop. (b) End to end Delay.

7. Conclusion

Vehicular ad-hoc networks are typically dynamic in terms of node mobility,

communication mechanism, network resource utilization, location management

etc. However, the end-user vehicular nodes are heterogeneous, which can range,

from high-end ARM processors to low-end PDAs or smart mobile devices.

Traditionally, middleware is required to abstract from this heterogeneity and to

enable the contextual network research challenges to focus on application issues.

The research work proposes to design and develop a QoS enabled middleware

service architecture that can deliver adaptive quality of network service for media

streaming applications between end to end nodes being engaged in session.

Providing handoff over VANET is well supported with HAND module which

adopts seamless transfer of media data using adaptive buffering technique based

on service in use. It was observed that HOSA provides consistent QoS throughout

the session as well maintains delay to be minimal. The aim is to identify solutions

for this realistic setting and to quantify the Quality of Service (QoS) being

supportable to user profile, service in use and network adopted. The work can be

extended with session manageable QoS such that variable sessions adopt

differential QoS architecture over the channel based call management approach.

References

1. Cespedes, S.; Shen, X.; and Lazo, C. (2011). IP mobility management for

vehicular communication networks: challenges and solutions. IEEE

Communications Magazine, 49(5), 187-194.

2. Johnson, D.; Perkins, C.; and Arkko, J. (2004). Mobility support in IPv6.

3. Gerla, M.; and Kleinrock, L. (2011). Vehicular networks and the future of the

mobile internet. Computer Networks, 55(2), 457-469.

4. Mohandas, B.K.; Nayak, A.; Naik, K.; and Goel, N. (2008). Absrp-a service

discovery approach for vehicular ad hoc networks. IEEE Asia-Pacific

Services Computing Conference, 1590-1594.



5. Wang, T.; and Wang, G. (2010). TIBCRPH: traffic infrastructure based

cluster routing protocol with handoff in VANET. Wireless and Optical

Communications Conference (WOCC), 1-5.

6. Roach, A.B. (2010). A SIP event package for subscribing to changes to an

HTTP resource.

7. Wu, W.; Banerjee, N.; Basu, K.; and Das, S.K. (2005). SIP-based vertical

handoff between WWANs and WLANs. IEEE Wireless Communications,

12(3), 66-72.

8. Sharma, M.; and Singh, G. (2012). Performance evaluation aodv, dymo, olsr

and zrp ad hoc routing protocol for IEEE 802.11 mac and 802.11 dcf in vanet

using qualnet. arXiv preprint arXiv:1202.1720.

9. Subramaniam, P.R.; Arunkumar, T.; and Sivanandam, S.N. (2011). A QoS

Adaptive Routing Scheme (QARS) for highly dynamic vehicular networks

with support to service and priority. World Applied Sciences Journal, 13(5),

1259-1268.

10. Kwak, D.; Mo, J.; and Kang, M. (2009). Investigation of handoffs for IEEE

802.11 networks in vehicular environment. IEEE First International

Conference on Ubiquitous and Future Networks, 89-94.

11. Indumathi, G.; and Murugesan, K. (2011). A cross-layer resource scheduling

with QoS guarantees using adaptive token bank fair queuing algorithm in

wireless networks. Journal of Engineering Science & Technology (JESTEC),

6(3), 260-267.

12. Mishra, A.; Shin, M.; and Arbaugh, W. (2003). An empirical analysis of the

IEEE 802.11 MAC layer handoff process. ACM SIGCOMM Computer

Communication Review, 33(2), 93-102.

13. Hosseini Seno, S.A. (2009). A distributed resource-management approach in

MANETs. Journal of Engineering Science & Technology, 4(2), 142-153.

14. Kaloxylos, A.; Lampropoulos, G.; Passas, N.; and Merakos, L. (2006). A

flexible handover mechanism for seamless service continuity in

heterogeneous environments. Computer Communications, 29(6), 717-729.

15. Van Hanh, N.; Ro, S.; and Ryu, J. (2008). Simplified fast handover in mobile

IPv6 networks. Computer Communications, 31(15), 3594-3603.

16. Leu, F.Y. (2009). A novel network mobility handoff scheme using SIP and

SCTP for multimedia applications. Journal of Network and Computer

Applications, 32(5), 1073-1091.

17. Márquez-Barja, J.; Calafate, C.T.; Cano, J.C.; and Manzoni, P. (2011). An

overview of vertical handover techniques: Algorithms, protocols and tools.

Computer Communications, 34(8), 985-997.

18. Bellavista, P.; Corradi, A.; and Foschini, L. (2007). Context-aware handoff middleware for transparent service continuity in wireless networks. Pervasive

and Mobile Computing, 3(4), 439-466.

19. Zhu, K.; Niyato, D.; Wang, P.; Hossain, E.; and In Kim, D. (2011). Mobility

and handoff management in vehicular networks: a survey. Wireless

Communications and Mobile Computing, 11(4), 459-476.



20. Musolesi, M.; and Mascolo, C. (2009). CAR: Context-aware adaptive routing

for delay-tolerant mobile networks. IEEE Transactions on Mobile

Computing, 8(2), 246-260.

21. Wang, X.; and Qian, H. (2013). A mobility handover scheme for IPv6-

based vehicular ad hoc networks. Wireless Personal Communications,

70(4), 1841-1857.

22. Kim, Y.D.; Moon, I.Y.; and Cho, S.J. (2009). A comparison of improved

AODV routing protocol based on IEEE 802.11 and IEEE 802.15. 4. Journal

of Engineering Science and Technology, 4(2), 132-141.

23. Awduche, D.G.T.L.V.S.D.; Berger, L.; Gan, D.; Li, T.; Srinivasan, V.; and Swallow, G. (2001). RSVP-TE: extensions to RSVP for LSP tunnels. RFC 3209.

24. Chang, Y.T.; Ding, J.W.; Ke, C.H.; and Chen, I.Y. (2010). A survey of

handoff schemes for vehicular ad-hoc networks. Proceedings of the 6th

International Wireless Communications and Mobile Computing Conference

1228-1231.

25. Project Kenai. (2011). JXTA. The language and platform independent protocol for P2P networking.

26. McCanne, S.; Floyd, S.; Fall, K.; and Varadhan, K. (1997). Network

simulator ns-2.

27. Tomandl, A.; Herrmann, D.; Fuchs, K.P.; Federrath, H.; and Scheuer, F.

(2014). VANETsim: An open source simulator for security and privacy

concepts in VANETs. IEEE International Conference on High Performance

Computing & Simulation, 543-550.

Documents

A CONTEXT AWARE BASED PRE-HANDOFF SUPPORT APPROACH …